Tableau 9.0 Beta, Some Network Graphs, And My Missions For This Year

Tableau 9.0 beta, some network graphs, and my missions for this year

View the interactive visualizations here.

The Internet Backbone, Europe, 2015.  Data via Maxmind.  Background map for reference purposes only (chord lines shown).

The Internet Backbone, 2005.  Data via The Opte Project.

Tableau 9.0 beta 

Tableau 9.0 beta has been released, and Tableau Public has shaped up to be real eye candy.  The UI in both desktop and Tableau Public have been given an overhaul, and there is a lot to like.  Regular expressions, random function, in pill editing, and the mapping gets a lot more power: in-map entity search, lasso/marquee select - although my favorite new capability (although the Apple Mac users have had it a while) - a color picker! With RGB and hex values!

Network graphs

I’ve been quietly sitting on the two data sets above, trying to make sense of the underlying fields; for example, with the maxmind dataset, I had to work out how to parse an interger as an octlet; i.e. 1.0.0.1 - and then using split_part in Postgre SQL (no CHARINDEX function available) to get rid of the end of block. The workbook has two inner joins and a union (Maxmind), and a second data source in MS Excel.

Why bother though?  Well, I’m fascinated by flow. Whether that is people, aircraft, money, or ip packets. Of course we can animate these graphs over time - but really, what are we looking for; what is the story?

Well I’ll tell you what I think know from experience what decision makers are looking for.  They are looking for anomalies. They want to know what are the biggest fluctuations, usually in the fattest “pipes”.  Flat lines and small pipes (think opportunities) are typically just noise.  For those building bar and pie charts into pretty looking dashboards, this is going to come as news and a bit of a shock.

Graph edges (vertices) can be encoded in size by metric - and once you build the matrix (array) you can also compute cost (distance is a normal consideration, time, bandwidth etc).  This can also be represented as a Sankey, but if there is a geo-spatial component to the data, then a map helps us interpret the data quicker.  As cave dwellers, we didn’t make bar or pie charts to tally up the hunt/gender ratios, we drew a map in the sands of how to get to the killing fields.

My Missions

I am going to continue pushing the integration of Business and Geo-spatial Intelligence.  I continue to champion Tableau, and I’m starting to integrate not just with ESRI, Mapbox, Geoserver etc but also with other XML/JSON based APIs, javascript libraries such as Cesium, d3, & three.js.

This is to support the next mission though - full sensory data exploration.   I’ll go further in depth at a later date; explain though some technologies, where I’ve got too so far, and why.